dunnhumby Media
Client: dunnhumby Media
Deliverables: dh Audiences application (v1.0)
Time: Sep 2019 - Mar 2020
Team: Product manager, UI designer, Engineering manager, 2 engineers
My role: User Research, UX/UI Design


dunnhumby Media tools help advertisers to grow Customer loyalty and enable retailers to monetise their own media. The retail media platform is powered by customer data science which includes many different targeting capabilities.

Problem space

The data science part of Media offerings includes different audience targeting approaches used in-house that are managed manually. In order to become more cost-efficient and provide faster services to clients, the process of creating targeted media campaigns should be automated.


Since there was no initial consensus for the project among different stakeholders, I first held a product vision workshop to collect all ideas and assumptions from all available stakeholders. We used product vision canvas to map out and align on the overall vision, target users and their needs, as well as to list product ideas and desired business goals. This was used to form assumptions and to guide us through the user interviews for the product discovery.

Product vision canvas Product vision canvas

Problems defined


Target audience

User research and ideation

With assumptions formed, we created an interview script aimed at in-house users, with questions that cover general media campaign process and current use targetings. In total, the team consisting of 2 designers and one product manager interviewed 35 people in 5 different roles inside the company. Common pain-points and needs were discovered by synthesizing the interview notes into affinity map. Based on this, we created a user journey for data analysts, who became our main user focus. For deeper understanding, we also mapped the process of media campaign creation to better understand the dynamics between different roles involved in daily work.

User journey Analyst user journey

User journey Campaign creation process flow

Key Findings

The following findings were discovered during the research:

Design and development

Together with product manager, we defined high level requirements to design specific scenarios, so we could validate some technical and UX assumptions about the features of the product. We designed a low fidelity prototype and ran usability tests with around 15 users in two iterations. The sessions were run remotely and conducted by us two designers and our product manager to ensure we always have two people attending the session. This helped to decide on the final set of requirements for the MVP, and enabled the UX team to start thinking about the UI design.

The design produced for the project relied on previously established global dunnhumby UI guidelines which I was a maintainer of, but we also introduced some new improvements such as new side navigation, step wizard and fluid grid layout. The development for the project relied on using the in-house component library which needed to be updated to accommodate all the changes that were required.

UI design - Dashboard

UI design - Audience details

UI design - Creating audience

UI design - Modal dialog UI design of Audiences product

Project outcomes

Check out my other case studies:

Back to projects